Math113:Introductionto

AbstractAlgebraSummer
2015
Announcements
Professor
Timeandplace
Coursecontentandprerequisites
Textbook
Exams,homeworkandgradingpolicy
Syllabus,readingandhomework
assignments

Announcements
(8/7)Pleasetakeafewminutestogive
yourCourseEvaluationforthisclass.
(8/6)AlongwithPS8,Ihavepostedan
outlineoftopicsandreviewexercisesforthe
finalexam.
(8/6)CorrectiononPS7:9.23shouldbe
9.22,asyouprobablyrealizedfromthe
hint.
(7/27)CorrectiononPS6:omitproblem
6.8.5(c).
(6/25)WenowhaveaabCoursespage.
YoucanloginwithyourCalNetIDtoview
yourscoresandhomeworksolutions.
(6/23)Correction:thefinalexamison
August13,thelastdayofclass,notAug14
asIpreviouslypostedbymistake.
(5/13)WelcometoMath113!Checkhere
forupdates.

https://math.berkeley.edu/~mhaiman/math113summer15/#text

1/8

Professor
MarkHaiman,
OfficehoursMonday,Tuesdayand
Wednesday121,orbyappointment.

Timeandplace
MondaythroughThursday1012,Room289
CoryHall.
Eachmeetingwillconsistofamixof
lectures,discussionandinclassproblem
solvingactivities.

Coursecontentandprerequisites
Thiscoursecoversthreemaintopics:(1)
thetheoryofgroups,withapplicationsto
symmetry(2)commutativerings,focusing
ontheringofintegersandpolynomialrings
and(3)fields,fieldextensions,andGalois
theory,withapplicationstogeometryand
algebra.
Theofficialprerequisiteforthiscourseis
Math54orequivalentpreparationinlinear
algebra.Itisalsohelpfultohaveprevious
experiencewithlogicandproofs,sets,and
mathematicalinduction,forinstanceas
coveredinMath55.
Youwillfindsomereviewmaterialonlogic,
sets,induction,complexnumbersandlinear
algebrainAppendicesAEofthecourse
textbook.

https://math.berkeley.edu/~mhaiman/math113summer15/#text

2/8

Textbook
FrederickM.Goodman,Algebra:Abstract
andConcrete,Edition2.6.
FollowthislinktodownloadthetextinPDF
format,freeofcharge.Theauthorrequests
thatyoumakeadonationtoacharitable
organizationofyourchoiceinlieuofa
royalty.

Exams,homeworkandgrading
policy
Gradeswillbebasedontwomidterm
exams,afinalexam,andweeklyhomework
assignments,accordingtotheformula:
Homework15%,Midterms25%each,Final
35%.
Examdates:
Midterm1July9
Midterm2July30
FinalExamAugust13
Allexamstakeplaceduringtheusualtwo
hourclassperiod.Youwillneedtobring
yourown`bluebooks'andscratchpaper.
Thefinalexamwillcoverallcoursematerial
butwithextraemphasisontopicsnotyet
coveredonmidtermexams.
Exampolicies:youmayconsulthandwritten
noteswhichyouprepareinadvance.No
otherbooks,printedmaterials,orelectronic
devicesmaybeused.
Ingeneraltherewillbenomakeup

https://math.berkeley.edu/~mhaiman/math113summer15/#text

3/8

examsandnodroppedscores.Incaseof
illnessoremergency,Iwillconsidermaking
allowancesformissedexamsonan
individualbasis,providedyouhaveavalid
reasonandcontactmeassoonaspossible.
Homeworkpolicies:usuallytwoorthree
problemsfromeachhomeworksetwillbe
chosenforgrading.
Youmayworkwithotherstosolve
homeworkproblems,butyoumustwrite
yoursolutionsindiviually.Copying
solutionsfromotherstudents,classesin
previousyears,theinternet,orelsewhereis
notallowed.
Youcanviewyourhomeworkandexam
scoresontheclassbCoursespage(choose
"Grades"fromthebCoursesmenu).Iwill
alsousebCoursestoposthomework
solutions.

Syllabus,readingandhomework
assignments
AllhomeworkisdueontheThursdayofthe
weekforwhichitisassigned.
Chapternumbersinreadingassignments
refertoGoodman.Ialsoplantodistribute
supplementarynotesonEuler'sfunction
(n),geometricconstructions,finitefields,
andtheGaloisgroupofageneralpolyomial.
Week

Topics

Reading Homework

Groupsof
symmetriesand

https://math.berkeley.edu/~mhaiman/math113summer15/#text

4/8

permutations.
June Divisibility,GCD,
1.11.7
2225 prime
factorization.
Modular
arithmetic.
Chinese
remainder
theorem.Euler's
function(n).
Groups,
subgroupsand
cyclicgroups.
June Dihedralgroups.
29
Homomorphisms
July2 andnormal
subgroups.
Cosets,
Lagrange's
theorem,order
andindexof
subgroups.
Quotientgroups.

July
69

Subgroupsof
G/NandZn.
Homomorphism
theorems.
Partitionsand
equivalence
relations.Direct
products.
Semidirect
products.Smith
normalformand
structureof
finitelygenerated
abeliangroups.
Symmetriesof
regularpolyhera.

https://math.berkeley.edu/~mhaiman/math113summer15/#text

Problem
Set1,
Solutions*

1.10,
2.12.5, Problem
2.7,
Set2,
Noteson Solutions
(n)

Problem
Set3,
2.62.7, Solutions.
3.1

Midterm1,
Solutions

3.2,3.5
3.6,4.1
4.3,5.1
5.4
Problem
5/8

July
Groupactions,
1316 orbits,
stabilizers
Burnside's
lemma.
Automorphisms,
classequation,
solvabilityofp
groups.

(through Set4,
Solutions
Cor.
5.4.5),
11.1,
11.4

Polynomials.
Ringsandfields.
Homomorphisms,
ideals,direct
sums,quotients.
Integral
domains,
relationship
1.8,
July
between
1.11,
2023
divisibilityand
6.16.4
principalideals,
PID's.Examples:
fields(familiar
andnew),
polynomialrings,
theringof
integers.
PIDimpliesUFD.
Primeand
maximalideals.
Constructionof
fieldsK[x]/(p(x))
wherep(x)isan
irreducible
polynomial.
Fractionfieldof
July
2730 anintegral
domain.
Divisibilityand

https://math.berkeley.edu/~mhaiman/math113summer15/#text

6.5,6.6
through
page
313,
6.8,7.1
7.2

Problem
Set5,
Solutions

Problem
Set6**,
Solutions.

Midterm2,
Solutions
6/8

gcdinaUFD.
Gauss'sLemma.
RUFDimplies
R[x]UFD.
Irreducibility
tests.

Aug
36

Vectorspaces
overafield.Field
extensions,
dimension,
algebraicand
transcendental
elements.Finite
andalgebraic
extensions.
Constructionof
extensionK()of
K.Rootsand
splittingfields.
Impossibilityof
angletrisection.
Finitefields.
PreviewofGalois
theoryfor
splittingfieldsin
thecomplex
numbers.

Separability
criteria.
Definitionand
characterization
ofGalois
extensions.Main
theoremof
Galoistheory.
Aug
Galoisgroupofa
1013 general
polynomial.
Solvablegroups,
https://math.berkeley.edu/~mhaiman/math113summer15/#text

Guideto
field
theory Problem
partsof Set7**,
3.3,7.3 Solutions
7.5,9.1
9.2

Noteson
finite
fields
partsof
9.39.7,
10.1
10.6

Problem
Set8,
Solutions

FinalExam,
Solutions

Online
7/8

simplegroups,
unsolvabilityof
thegeneral
quinticby
radicals.

Review
Guide

Course
Evaluation

*Homeworkproblemschosenforgrading:

ProblemSet1:1.4.2(5pts),1.6.3(3
pts),1.6.12(a)(2pts)
ProblemSet2:2.1.2(3pts),2.4.4(4
pts),2.5.8(3pts)
ProblemSet3:2.5.14(3pts),2.7.6(5
pts),3.1.10(2pts)
ProblemSet4:3.5.1(3pts),3.6.13(5
pts),5.1.5(2pts)
ProblemSet5:6.2.2,6.2.14(5each)
ProblemSet6:6.4.14(4pts),6.6.34
(6ptscombined)
ProblemSet7:7.4.6(5pts),9.1.4(5
pts)
ProblemSet8:Finitefieldsproblem(1
or2ptsperpart)
**Corrections:PS6omit6.8.5(c)PS7

9.23shouldbe9.22.

Backtotop|Prof.Haiman'shomepage|
Calendar

https://math.berkeley.edu/~mhaiman/math113summer15/#text

8/8

Math 113, Summer 2015

Prof. Haiman
Notes on Eulers function (n)

For each positive integer n, Eulers function (n) is defined to be the number of positive
integers k less than n which are relatively prime to n.
For example, of the positive integers less than 12, four are relatively prime to 12, namely
1, 5, 7, and 11. Therefore (12) = 4.
The purpose of these notes is to discuss some properties of (n). The same topics are
covered in Section 1.9 of Goodmans book, but I prefer a different and I think somewhat
simpler approach.
Before reading these notes, you will need to read Sections 1.6 and 1.7 of Goodman. I
will use the same notation as he does for congruence, residue classes, and the system Zn of
residue classes, with its operations of addition and multiplication modulo n. We will make
use of the Chinese Remainder Theorem, which is Proposition 1.7.9 in Goodman.
1. Multiplicative inverses in Zn
Recall that each residue class [a] in Zn has a unique representative with a in the range
0 a < n. We will begin by showing that the classes [a] which have a multiplicative inverse
in Zn are exactly those for which a is relatively prime to n (this is Proposition 1.9.9 in
Goodman).
First, suppose a is relatively prime to n. Since a and n are relatively prime, there are
integers s and t such that 1 = sa + tn. Then sa 1 (mod n), which means [s][a] = [1] in
Zn , so [s] is the required inverse.
For the converse, suppose a is not relatively prime to n. Let d = gcd(n, a). Then d > 1,
so l = n/d is a positive integer less than n, and therefore [l] 6= [0] in Zn . Now la = n(a/d) is
a multiple of n, since d divides a, so [l][a] = [0] in Zn . If [a] had a multiplicative inverse [b]
we could multiply on both sides by [b] to get [l] = [0] in Zn , a contradiction.
I will use the notation Z
n for the set of residue classes [a] in Zn which have multiplicative
inverses. We have just seen that Z
n consists of those classes [a] for which a is relatively
prime to n. The cardinality of the set Z
n is therefore equal to the number of integers a in
the range 0 a < n which are relatively prime to n. But 0 is not relatively prime to n
(why not?), so the cardinality of Z
n is the number of positive integers less than n which are
relatively prime to n. In other words, (n) = |Z
n |. This fact is the reason why the function
(n) is important.
2. A formula for (n)
Theorem. Let the prime factorization of n be n = pe11 pekk . Then
(1)

(n) =

k
Y

pei i 1 (pi 1)

i=1

Example: the prime factorization of 12 is 22 3. According the formula in the theorem,

we have (n) = 21 (2 1) 30 (3 1) = 4, in agreement with what we found before.
1

We will prove (1) in two steps. First, we will show that (n) = pe1 (p 1) if n = pe is a
power of a prime.
Second, we will use the Chinese Remainder Theorem to show that if m and n are relatively
prime, then (mn) = (m)(n). This implies (by induction on k) that if m1 , . . . , mk are
pairwise relatively prime, then (m1 mk ) = (m1 ) (mk ).
Formula (1) will then follow, because if n = pe11 pekk , then the factors mi = pei i are
pairwise relatively prime, and (mi ) = (pei i ) = pei i 1 (pi 1).
Now let us consider the case n = pe . Since p is the only prime factor of n, a number a is
relatively prime to n if and only if p does not divide a. There are pe integers a in the range
0 a < pe . Of these, pe1 are multiples of p, namely the numbers rp for 0 r < pe1 . This
leaves pe pe1 = pe1 (p 1) integers 0 a < n relatively prime to n, and they are all
positive, since a = 0 was one of those excluded. This shows that (n) = pe1 (p 1).
It remains to show that if m and n are relatively prime, then (mn) = (m)(n). An
integer x is relatively prime to both m and n if and only if x has no prime factor in common
with either m or n, if and only if x has no prime factor in common with mn. So x is relatively
prime to both m and n if and only if x is relatively prime to mn (this much is true even if
m and n are not relatively prime).
Since we are dealing with more than one modulus at the same time, I will write [x]m ,
[x]n , or [x]mn to distinguish between residue classes in Zm , Zn , or Zmn . Since m and n
are relatively prime, the Chinese Remainder Theorem gives a one-to-one correspondence
between residue classes [x]mn in Zmn and pairs ([a]m , [b]n ), with [a]m Zm and [b]n Zn . In
the direction from Zmn to Zm Zn , the correspondence simply sends [x]mn to ([x]m , [x]n ).
We have just seen that x is relatively prime to mn if and only if it is relatively prime to
both m and n. Therefore, in the correspondence given by the Chinese Remainder Theorem,

Z
mn corresponds to Zm Zn . This shows that |Zmn | = |Zm ||Zn |, so (mn) = (m)(n). 
The theorem above is equivalent to Goodman, Proposition 1.9.18(a), although Goodman
expresses the formula a bit differently. Goodmans Proposition 1.9.18(b) is what we proved
in the second part of the proof given above.
3. Eulers theorem
Theorem. If a is relatively prime to n, then a(n) 1 (mod n).
This is Theorem 1.9.20 in Goodman. He outlines a fairly complicated proof in the exercises
to Section 1.9. At the end of Section 1.10 he goes on to explain how it can be deduced more
easily from a general theorem of group theory. I will just add a few comments on the
explanation Goodman gives in 1.10.
Goodman uses the notation (n) for the set of residue classes in Zn which have multiplicative inverses, which I denoted Z
n . We have seen that this is also the set of classes of
integers relatively prime to n, and therefore that |Z
n | = (n).
Now if [a] and [b] in Zn have multiplicative inverses, then [b]1 [a]1 is an inverse of [a][b], as
you can check. This shows that the subset Z
n is closed under the operation of multiplication
in Zn . It also contains the multiplicative identity [1] (which is its own inverse). Multiplication

is associative in Zn and therefore also in Z

n . Therefore, since in Zn we have the identity

and inverses, Zn is a group with the operation of multiplication (this is Goodman, Lemma
1.10.3).

Now we invoke the general theorem (Goodman, Theorem 2.5.6, which we will prove later)
that every element a in a finite group of cardinality g satisfies ag = e, where e is the identity
(n)
element. When the group is Z
= [1], which is another way of writing
n , this becomes [a]
Eulers theorem.

Math 113, Summer 2015

Prof. Haiman
A guide to field theory

For the last part of the course we will be studying fields and some classical applications
of field theory, such as to prove that that it is not possible to trisect an arbitrary angle with
straightedge and compass, and that there is no formula for the roots of a polynomial of degree
five or more, analogous to the familiar formula for the roots of a quadratic polynomial.
Most of the theory that we will need is covered in Goodmans text, although I will add a
few things, mainly about finite fields (see the separate set of notes on this subject) and the
impossibility of angle trisection. Goodman treats the subject more comprehensively than
we will have need or time for. In these notes, I will outline the parts of the subject that I
would like you to learn, and where to find them in Goodman.
Here and there I will also mention simplifications or alternative approaches to what you
will find in the text.
1. Vector spaces
All the familiar operations of vector and matrix algebra, such as solving a system of linear
equations A x = 0 by row-reducing the matrix A, work for matrices and vectors with entries
in any field K.
We do require that K be a field, and not just a commutative ring with identity, because
it is essential to be able to divide by non-zero scalars. For example, just to solve a single
linear equation a x = b in one variable x, you need to be able to divide by a.
For a review of vector and matrix algebra you might want to read Goodman Appendix
E.1-2.
The definition of an abstract vector space V over any field K and some fundamental
examples are given in Goodman 3.3.1-3 and 3.3.6-7.
The definition and properties of a basis of V , and the dimension dim(V ) if V is finitedimensional, are in Goodman 3.3.15-28.
You will notice that Goodman proves Proposition 3.3.25 using the fact that if C is a matrix
with more columns than rows, then the vector equation C a = 0 has non-zero solutions a.
This should be familiar to you in the case when K is the field of real numbers, but it is
equally valid for other fields.
One important point which Goodman skips over is that a subset S of V is a basis if and
only if every element of V can be expressed uniquely as a linear combination of elements of
S. Exercise: prove this.
The above is all that we will need from linear algebra.
2. Field extensions
Carefully read all of Goodman 7.3 for the definition and basic properties of field extensions
K L. We will be studying the theory of finite extensions, which are always algebraic.
The dimension dimK (L) and the formula dimK (M ) = dimK (L) dimL (M ) for K L M
(Goodman 7.3.1) will be important for us.
1

The minimal polynomial f (x) K[x] of an element L, and the description of the
subfield K() L, isomorphic to K[x]/(f (x)), will also be especially important (Goodman
7.3.5-8).
3. Solution of the cubic, and splitting fields of polynomials in C
In 7.2, Goodman explains how to solve a cubic equation, after reducing it to the special
form x3 + px + q = 0.
In 7.4, he works out the description of field extensions K L, where K is a subfield of C
containing the coefficients of a cubic f (x) = x3 + px + q = 0, and L = K(1 , 2 , 3 ) is the
extension of K generated by the three complex roots i of f (x), called the splitting field of
f (x) over K.
In 7.5, he outlines the general picture for the splitting field in C of a polynomial f (x) with
coefficients in a subfield K of C.
In class I will go over the example of the cubic equation, as discussed in Goodman 7.2 and
7.4, along with other examples, to motivate and illustrate the theory we will be developing.
You should read 7.2 and 7.4 in full.
You can skip all proofs and many of the intermediate results in 7.5. It will be enough
to understand the statements of Theorem 7.5.1, 7.5.7-9, and 7.5.11, and how they apply in
examples 7.5.12-14 and any others we might discuss in class.
4. Impossibility of angle trisection
This topic is not in Goodman, so I will explain it briefly here (and in class). We will only
need the material in Goodman 7.3 for this.
We define a real number to be constructible if it is the x or y coordinate of a point in
the plane constructible by straightedge and compass, starting with nothing but two marked
points: one at the origin (0, 0) and one on the x axis at (1, 0), which serve to establish a unit
of length.
For example, since we can construct
a unit square, and mark off a segment on the x-axis
congruent to its diagonal, the number 2 is constructible.
There are only three ways to construct new points: intersect two lines, a line and a circle,
or two circles. The only lines we can construct are those passing through two constructible
points, and the only circles are those centered on a constructible point and having constructible radius. You can verify, by working out the equations for the intersection points of
two lines, or a line and a circle, or two circles, that every new number produced by these constructions is a solution of a quadratic equation whose coefficients are previously constructed
real numbers.
This shows that every constructible real number is contained in a finite iterated quadratic
extension of Q, that is, a subfield K R such that there is a tower of intermediate fields
Q = K0 K1 Km = K
in which each Ki+1 has the form Ki (i ), where i is a root of a quadratic equation with
coefficients in Ki . Then dimKi (Ki+1 ) = 2 for all i, so dimQ (K) = 2m . If is a constructible
real number, then Q() is a subfield of a field K as above. Hence dimQ (Q()), which is the
degree of the minimal polynomial of over Q, divides 2m , and so is itself a power of 2.

(It can be shown, conversely, that all elements of any iterated quadratic extension of Q in
R are constructible. For this one has to exhibit geometric constructions for the arithmetic
operations and square roots. We wont need this.)
Since we can construct an equilateral triangle, we can construct two lines meeting at at
angle of 2/3 = 120 . If there were a construction to trisect an angle, we could then use
it to construct an angle of 2/9, which would imply that the real number 2 cos 2/9 is
constructible.
To prove that angle trisection is impossible, it will therefore suffice to show that the
minimial polynomial of 2 cos 2/9 over Q has degree 3.
To this end, let = e2i/9 , so that 2 cos 2/9 = + 1 . Since 9 = 1, we see that is a
root of x9 1 = 0. The latter polynomial factors as
x9 1 = (x3 1)(x6 + x3 + 1).
Since 3 6= 1, must be a root of the second factor, so it satisfies 6 + 3 + 1 = 0. Dividing
by 3 , we get
3 + 3 + 1 = 0,
or
( + 1 )3 3( + 1 ) + 1 = 0.
This shows that 2 cos 2/9 = + 1 is a root of the polynomial x3 3x + 1. To complete
our argument we need only show that this last polynomial is irreducible in Q[x]. Since it
is a cubic polynomial, it suffices to show that it has no rational root. By the rational root
test, the only possible rational roots are 1. But neither of these is a root, so x3 3x + 1 is
irreducible.
Exercise: Show that x3 3x + 1 has three real roots, equal to 2 cos 2/9, 2 cos 4/9 and
2 cos 8/9.
5. Splitting fields, automorphisms, and Galois theory
In Goodman Chapter 7 we encountered the notion of the splitting field in C of a polynomial
f (x) with coefficients in a subfield K of C. These concepts apply in a more general setting.
Given any field K and a polynomial f (x) K[x], we can prove that there is an algebraic
extension K L such that f (x) factors into linear factors in L[x]that is, L contains a
complete set of roots of f (x)and the roots of f in L generate L as extension of K. This
extension L is called the splitting field of f (x) over K.
Sections 9.1-5 of Goodman discuss the theory of splitting fields and their automorphism
groups, leading to the fundamental theorem of Galois theory (Goodman 9.5.4).
We will need much but not all of the material in these sections.
In 9.1, we only need Proposition 9.1.1.
Section 9.2 discusses the existence of splitting fields and how to construct them, their
uniqueness up to isomorphism, and the construction of specific isomorphisms and automorphisms in appropriate circumstances. Everything in 9.2 is important for us.
Section 9.3 discusses criteria for a polynomial to have multiple roots. You may already
know that a polynomial f (x) with real coefficients has distinct roots (including complex
roots) if and only if f (x) and its derivative f 0 (x) are relatively prime.

In Goodman Exercises 9.3.1-5 you will work out the corresponding criterion for polynomials over any field. (The phrase if Df (x) is not identically zero, then in 9.3.5 could be
omitted, since if Df (x) = 0, then the gcd of f (x) and Df (x) is f (x).)
The rest of 9.3 discusses some applications of the criterion in Exercise 9.3.5. The main
point is that if K has characteristic zero, or if K is finite, then an irreducible polynomial
over K always has distinct roots in any extension of K. This includes most fields K that
will be of interest to us.
Section 9.4 discusses the automorphism group AutK (L) of a splitting field K L. We
will need everything in this section except maybe Corollaries 9.4.18-19.
There is a mistake in Goodmans proof of Proposition 9.4.1: it should refer to Proposition
9.2.4 instead of Corollary 9.2.5. Or, better, study the proof of Proposition 9.4.2 first, and then
notice that Proposition 9.4.1 is the special case of 9.4.2 when M = K() and M 0 = K().
The most subtle, but also the most fundamental, result in 9.4 is Theorem 9.4.13. Here is
a slightly different proof which you might find simpler and more conceptual.
The first part is the same as in Goodman: we list the roots 1 , . . . , r of f (x) in L and
consider the tower of subfields
K = M0 M1 Mr = L
in which Mi = Mi1 (i ). We will prove that if is a root of f (x) in L, then for any
subfields K M M () L, every element of M () fixed by AutK (L) belongs to M . It
follows that if Fix(AutK (L)) Mi+1 , then Fix(AutK (L)) Mi . Since we obviously have
Fix(AutK (L)) Mr , we can use this repeatedly to conclude that Fix(AutK (L)) M0 . In
other words, Fix(AutK (L)) = K, which is what we wanted to prove.
The second part is to prove that every element of M () fixed by AutK (L) belongs to M .
I will do this a bit differently than Goodman does. Since AutM (L) AutK (L), it is enough
to prove that every element of M () fixed by AutM (L) belongs to M . In other words, the
problem really only concerns the extension M M () L and the group AutM (L).
Let N = M ()Fix(AutM (L)). We have M N M () and want to show that M = N .
Let p(x) be the minimal polynomial of over M , and q(x) its minimal polynomial over N .
Since is a root of f (x), p(x) is an irreducible factor of f (x) in M [x], so p(x) is separable.
Let l = deg(p(x)) = dimM (M ()). Since L is the splitting field of f (x) over M as well as over
K, p(x) has l distinct roots 1 , . . . , l in L, one of which is . By 9.4.4 (b) (applied to M L
rather than to K L), AutM (L) acts transitively on the i . Since q(x) has coefficients fixed
by AutM (L) and is a root of q(x), every i is a root of q(x). Hence the degree of q(x),
which is equal to dimN (M ()), is at least l. In other words, dimN (M ()) dimM (M ()).
But since dimM (M ()) = dimM (N ) dimN (M ()), this implies that dimM (N ) = 1, that is,
M = N.
In 9.5, we only need 9.5.1-4. Proposition 9.5.1 is known as the Primitive Element Theorem.
It is used here to prove Proposition 9.5.3, which in turn is used, together with results from
Section 9.4, to prove Theorem 9.5.4.
The main results in Galois theory are Theorem 9.4.15, Corollary 9.4.16, Proposition 9.4.17,
and Theorem 9.5.4.
6. Unsolvability of the quintic equation

I am more interested in having you appreciate this topic as a demonstration of the power
of Galois theory than in your following every technical detail. For this reason, I will first
give you an outline of the general ideas before pointing you to the specifics in Goodman.
It will be helpful to begin by thinking about the familiar formula

b b2 4ac
2a
for the roots of a quadratic equation
a x2 + b x + c = 0.
One way to view the symbols a, b and c here is that they stand for complex numbers, which
we specify each time we want to use the formula. To address the question of whether such
formulas exist for higher degree polynomials, however, it is more useful to take a different
point of view, and regard the symblols a, b, c as abstract indeterminates, like the symbol x
in a polynomial p(x).
In other words, we consider the field of rational functions K = C(a, b, c), the fraction field
of the polynomial ring C[a, b, c] in three variables. Then f (x) = ax2 + bx + c is a polynomial
over K, that is, an element of K[x]. The expression = b2 4ac is an element of K, but it
is not the square of an element of K. Thus p(z) = z 2 has no root in K, and is therefore
irreducible (since it has degree 2).
We can introduce a square root of by forming the
extension E = K[z]/(p(z)) = K( ). The formula

b b2 4ac
b
=
2a
2a
now stands for either of two elements (depending on the sign) 1 , 2 E, which are, of
course, the rootsof f (x) in E.
Thus E = K( ) is the splitting field of the generic quadratic polynomial f (x) = ax2 +
bx + c over K = C(a, b, c). Splitting fields always exist, but in this case, the quadratic
formula allowed us to construct it in a special way: namely, by adjoining a square root of
the element K.
We can interpret Cardanos formulas for the roots of a cubic equation, worked out in
Goodman 7.2, in a similar manner. We might take the generic cubic to be ax3 + bx2 + cx + d,
but for the problem of finding the roots we can always simplify first, as in Goodman 7.2, to
the form
f (x) = x3 + p x + q.
We now regard this as a polynomial over the field of rational functions K = C(p, q). Note
that this is quite different from the point of view taken in Goodman 7.4, where p and q were
complex numbers and K was a subfield of C containing them. Now C is a subfield of K
instead!
In Cardanos formulas, we first introduce the square root

q 2 p3
, where =
+ .
4
27
Then we introduce a cube root

q
3
A = B, where B = + .
2

After this, the roots of f (x) are given, as in Goodman 7.2, by

p
p
p
, 2 = A
, 3 = 2 A
,
1 = A
3A
3 A
3 2A
where = e2i/3 (note that our field K contains
the complex number to begin with).

3
In other words, the field E = K( )( B) is the splitting field of the generic cubic
f (x) = x3 + px + q over K = C(p, q). Again, the splitting field always exists, but Cardanos
formulas have allowed us to construct it by adjoining first a square root and then a cube
root to K.
A formula for the roots of a polynomial equation involving only arithmetic operations,
radicals (meaning n-th roots), and perhaps some specific complex numbers such as the
in Cardanos formulas is called a solution of the equation by radicals. Besides the familiar
formula for the quadratic equation and Cardanos formulas for the cubic equation, there is
also a known solution of the general fourth degree equation by radicals.
We shall use Galois theory to prove that there is, however, no solution by radicals of the
general quintic (degree 5) equation
x5 + a x4 + b x3 + c x2 + d x + e = 0
(which we have taken to be monic for simplicity and without loss of generality). In the
process we will also see the natural explanation for the fact that 5 is the smallest degree in
which no solution by radicals exists.
From our discussion so far we already know how to begin. Take K = C(a, b, c, d, e), let
f (x) K[x] be the generic quintic
f (x) = x5 + a x4 + b x3 + c x2 + d x + e,
and let E be the splitting field of f (x) over K. If the quintic were solvable by radicals, it
would mean that E, or maybe some larger extension K E L, can be constructed from
K in stages, where at each stage we adjoin an n-th root of some element. In other words,
we would have a tower of extensions
(1)

K = K0 K1 Kr = L,

with Ki+1 = Ki ( ni i ) for some i Ki and positive integer ni , for each i. We have only
required E L, rather than E = L, to allow for the possibility that there might be a
solution by radicals which gives some extraneous solutions in addition to the roots of f (x).
Our conclusions will imply that not even this is possible.
The steps to complete the proof are as follows. By definition, E is the splitting field of
a polynomial f (x) over K. Since our fields have characteristic zero, f (x) is separable, so
K E is a Galois extension.
The first thing to prove is that the Galois group AutK (E) is the full permutation group
S5 acting on the five roots of f (x) in E. More generally, the Galois group of any generic
polynomial of degree n is the full permutation group Sn of its n roots in a splitting field.
Goodman proves this in 9.6 and 9.7 (Theorem 9.7.1), but there is an easier way, which
I will explain briefly. The idea is to prove by induction that dimK (E) = n!. Then, since
| AutK (E)| = dimK (E) and AutK (E) is a subgroup of Sn , it will follow that AutK (E) = Sn .

The degree n generic polynomial

f (x) = xn + an1 xn1 + + a1 x + a0
is irreducible in C[a0 , . . . , an1 , x] since it is linear in a0 . By Gausss Lemma, it follows that
f (x) is irreducible over K = C(a0 , . . . , an1 ). Adjoining a root of f (x) to K therefore gives
an extension K K() with dimK (K()) = n. The polynomial f (x) factors over K() as
f (x) = (x )g(x), for a polynomial g(x) = xn1 + bn2 xn2 + + b1 x + b0 with coefficients
bi K(). Our splitting field E is also the splitting field of g(x) over K(). If we verify that
g(x) is again a generic polynomial, then we will have dimK() (E) = (n 1)! by induction,
and therefore dimK (E) = dimK (K()) dimK() (E) = n (n 1)! = n!.
To prove that g(x) is generic we need to show that K() is isomorphic to a field of rational
functions C(b0 , . . . , bn1 , ). We can construct this isomorphism by working out formulas for
the bi in terms of the ai and vice versa, as I will
do in class.

n
Next we observe that
if M
contains C,and is an n-th root of some M , then allthe
n-th roots of are n , n , . . . , n1 n , where = e2i/n . The extension M M ( n )
n
is
thereforethe splitting field
over M , so it is Galois. If gk AutM (M ( n )) sends

of x j+k
k n
j n
n
n
to , then
gk ( ) =
. This gives gk gj = gj+k and shows that the Galois

n
group AutM (M ( )) is abelian.

A more general discussion of radical extensions M M ( n ) can be found in Goodman

10.4-5, but we will not need it. The simpler observations in the paragraph above will do.
The remaining steps are the same as in Goodman 10.1-3 and 10.6.
The extension K L is not necessarily Galois, but we can always find a larger extension
K L L0 which is Galois, and is still given by a tower of radical extensions as in (1).
This is Goodman,Lemma 10.6.3. The basic idea is that every time we want to construct
an extension M ( n ), we should not just adjoin an n-th root of , but of every root of
the minimal polynomial p(x) of over K, which gives the splitting field of the polynomial
p(xn ) K[x] over M . The resulting extension will then be Galois over K if M was.
Replacing L with L0 , we can now assume that L is Galois over K. Using the fundamental
theorem of Galois theory and the tower of intermediate fields (1), we conclude as in Goodman
10.6.4 that the Galois group G = AutK (L) contains a tower of subgroups
{e} = G0 G1 Gr = G
with each Gi normal in Gi+1 and Gi+1 /Gi abelian for each i. Such a group G is said
to be solvable (Goodman, Definition 10.1.4). Our original splitting field E of f (x) is an
intermediate extension K E L, Galois over K. This implies that its Galois group
AutK (E) is a quotient group G/N of G.
In class we will prove Goodman, Exercise 10.2.6: every quotient of a solvable group is
solvable. We conclude that if f (x) is solvable by radicals, then the Galois group AutK (E)
of its splitting field must be a solvable group. This is the origin of the term solvable for
groups.
To complete the proof that the general quintic is not solvable by radicals, it only remains
to show that the symmetric group S5 is not solvable. This follows because (i) the only normal
subgroups of S5 are {e}, A5 and S5 , and (ii) A5 is non-abelian and simple (has no proper
non-trivial normal subgroup). This is Goodman, 10.3.2, 10.3.4, and Exercise 10.3.6.

A final remark: S2 , S3 and S4 are solvable (A4 is non-abelian, but not simple because it
has a normal subgroup with four elements). This explains why there are formulas for the
solution by radicals of general polynomial equations of degrees 2, 3 and 4, but not 5.

Math 113, Summer 2015

Prof. Haiman
Notes on finite fields

1. The order of a finite field

Recall (Goodman 6.4.9) that the subring generated by 1 in any integral domain R is
isomorphic either to Z, in which case we say R has characteristic zero, or to Zp , in which
case we say R has characteristic p. If F is a field of characteristic zero, then F is clearly
infinite. In fact, since F is a field, it not only contains a copy of Z, but a copy of the fraction
field Q of Z.
A finite field F must therefore have characteristic p for some prime p, that is, the subring
of F generated by 1 is isomorphic to Zp . Note that this subring is already a subfield. We
can identify it with Zp and think of Zp F as a field extension.
In particular, F is a vector space over Zp , and since F is finite, d = dimZp (F ) is finite.
Then F is isomorphic as a vector space (and as an abelian group, but not as a ring!) to
(Zp )d . Hence F has pd elements.
Our main goal in these notes will be to prove
Theorem 1.
(i) For every prime power q = pd , there exists a finite field F(q) of order q.
(ii) F(q) is unique up to isomorphism.
(iii) F(q) can be constructed as Zp (), where is a root of an irreducible polynomial f (x)
of degree d in Zp [x].
In the process we will also learn something about the structure of the finite fields F(q),
and use this knowledge to discover an algorithm for testing whether a polynomial f (x) over
Zp is irreducible in Zp [x].
2. The Frobenius automorphism
Proposition (Goodman 9.3.3). If F is a field of characteristic p, the map : F F
given by (x) = xp , called the Frobenius homomorphism, is a ring homomorphism. The
Frobenius homomorphism is always injective. If F is finite, then is bijective, that is, it is
an automorphism.
Proof. It is clear that (xy) = xp y p = (x)(y). We also need to prove that (x + y) =
(x) + (y). By the binomial theorem,
p  
X
p k pk
p
(1)
(x + y) = (x + y) =
x y .
k
k=0
Recall that

 
p
p!
=
.
k
k!(p k)!
For 0 < k < p, k! and (p k)! are products of positive integers less than p. Hence p does not
divide the
denominator in the above fraction. Since p divides the numerator, we see that p
divides kp . Bearing in mind that p z = 0 for every element z in a field of characteristic p,
1

we see that the terms for 0 < k < p on the right hand side in (1) are all zero. The remaining
terms, for k = 0 and k = p, are xp and y p . This gives
(x + y) = (x + y)p = xp + y p = (x) + (y).
We have now shown that is a ring homomorphism. It is not zero, since (1) = 1, so
its kernel is an ideal I F , I 6= F . But since F is a field, the only such ideal is I = {0}.
Hence is injective. (This argument actually shows that every unital ring homomorphism
: F R from a field to any ring with identity is injective.)
If F is finite, then , being an injective map from F to F , is also surjective.

We will now prove part (i) of Theorem 1, that for every prime power q = pd , a finite field
of order q exists.
Given q = pd , let F be the splitting field (Goodman 9.2.3) over Zp of the polynomial
P (x) = xq x in Zp [x]. Since p divides q, the formal derivative of P (x) is P 0 (x) = 1, which
is (obviously) relatively prime to P (x). By the derivative criterion (Goodman 9.3.5), P (x)
has no multiple roots in any extension field of Zp . In particular, P (x) has q distinct roots in
its splitting field F .
d
d
For an element F to be a root of P (x) means that p = , or, since p = d (),
that the d-th power d of the Frobenius automorphism fixes .
Since F is generated by roots of P (x), this implies that d fixes every element of F . In
other words, every element of F is a root of P (x). Since P (x) has q roots in F , this shows
that |F | = q.
Now we prove part (ii) of Theorem 1, that all finite fields of order q are isomorphic. We
know (Goodman 9.2.5) that the splitting field of P (x) over Zp is unique up to isomorphism,
but we still need to show that if E is another field of order q, then E is a splitting field for
P (x).
So, suppose |E| = q, without assuming in advance that E is a splitting field for P (x). The
multiplicative group E = E \ {0} has order q 1, so by Lagranges Theorem, every x E
satisfies xq1 = 1, and consequently xq = x. But of course x = 0 also satisfies xq = x. This
shows that every element of E is a root of P (x) = xq x. Since |E| = q, it follows that E
is a splitting field for P (x).
From now on we write F(q) for the splitting field of P (x), which we have just shown is the
unique finite field of order q, up to isomorphism.
To prove part (iii) of Theorem 1, we just have to show that F(q) can be generated over
Zp by a single element . Then by the basic theory of field extensions, we have F(q) =
Zp ()
= Zp [x]/(f (x)), where f (x) Zp [x] is the minimal polyonomial of , which will be a
polynomial of degree d = dimZp (F(q)).
It follows from the structure theorem for finite abelian groups that the multiplicative group
F of any finite field is cyclic. This is shown in Goodman, Theorem 3.6.25. Ill remind you
what the essential point there is. Since F is a finite abelian group, it has an invariant factor
decomposition F
= Zn1 Zn2 Znk , where each ni divides the one before. Then
every element x F satisfies xn1 = 1. However, since F is a field, the equation xn1 1 = 0
cannot have more than n1 roots, giving |F | n1 . But |F | = n1 nk , so this implies that
F has just one factor in its invariant factor decomposition, that is, F is cyclic.

Now let F be a generator of F as a cyclic group. Then also generates F as an

extension of Zp .
Just to be clear, I should point out that the above is just one possible way to find a
generator of F(q) over Zp . There are often other elements such that F(q) = Zp (), but
does not generate the group F(q) . For example, in F(9), we have F(9)
= Z8 , which
has four elements that generate it as a cyclic group. But since dimZ3 F(9) = 2, the only
subfields of F(9) are itself and Z3 = F(3). Hence any element F(9) which is not in Z3 is
a generator. There are six such elements, but only four of them are generators of F(9) .
3. Extensions of finite fields
Let us now work out for which q and r there can be an extension of finite fields F(r) F(q).
Of course both fields must have the same characteristic, so q and r must be powers of the
same prime, say q = pd and r = pe . Also, since dimF(r) (F(q)) = dimZp (F(q))/ dimZp (F(r)) =
d/e, we must have e dividing d.
We will now prove that these conditions are sufficient, that is, if e divides d then F(pd ) has
a subfield E of order pe , and moreover this subfield is unique. (We know that E is unique up
to isomorphism, being isomorphic to F(pe ), but that is not sufficient to conclude that F(pd )
has only one such subfield E.)
d
e
For this we consider the polynomials P (x) = xq x = xp x and Q(x) = xr x = xp x
in Zp [x]. We will show that if e divides d, then Q(x) divides P (x), or in other words,
xq x belongs to the ideal (xr x) Zp [x]. Let d = k e, so q = rk . In the quotient ring
2
3
2
Zp [x]/(xr x) we have xr x and therefore xr = (xr )r xr x, xr = (xr )r xr x,
and so on. In particular, xq x, which means that xq x (xr x).
Now, since F(q) is a splitting field of P (x), and Q(x) is a factor of P (x), F(q) contains
r roots of Q(x), that is, it contains a splitting field E of Q(x), which we have already seen
is isomorphic to F(r). Furthermore, any subfield E 0 F(q) of order r is a splitting field
of Q(x) and therefore contains all the roots of Q(x) in F(q). In other words, E E 0 , and
therefore E = E 0 since |E| = |E 0 | = r. This shows that E is unique.
Looking ahead a bit, the picture we have just worked out can be understood nicely in
terms of Galois theory. Since F(q) is the splitting field of the separable polynomial P (x)
over Zp , the extension Zp F(q) is a Galois extension.
The Frobenius automorphism is an element of the Galois group G of F(q) over Zp . Its
fixed field consists of the roots of the equation xp x = 0 in F(q). But this equation has
only p roots, so the fixed field of , or of the cyclic subgroup hi G, is just Zp . By the
Galois correspondence, this implies that G = hi.
In other words, the Galois group G of F(q) over Zp is cyclic of order d (where q = pd ),
and generated by . Now G
= Zd has one subgroup for each divisor e of d, namely the
cyclic subgroup generated by e . These subgroups are in one-to-one correspondence with
the subfields of F(q): specifically, the fixed field of the subgroup he i is the unique subfield
E F(q) of order pe .
4. Irreducibility of polynomials over Zp
Part (iii) of Theorem 1 implies that there exist irreducible polynomials in Zp of every
degree d > 0. Actually, we can say much more:

Proposition. For q = pd , the polynomial P (x) = xq x is exactly the product of all monic
irreducible polynomials f (x) in Zp [x] of degree dividing d.
Proof. Since P (x) does not have repeated roots, it is a product of distinct irreducible factors,
which we can take to be monic, since P (x) is monic. Since the roots of P (x) in its splitting
field F(q) are all the elements of F(q), the irreducible factors are precisely the minimal
polynomials of elements of F(q). In particular, their degrees are the dimensions over Zp of
subfields E F(q), so they divide d.
Conversely, if f (x) Zp is irreducible of degree e dividing d, then it has a root in F(pe )
=
e
Zp [x]/(f (x)). We saw in the previous section that that F(p ) is isomorphic to a subfield of
F(q), so f (x) has a root in F(q), and is therefore an irreducible factor of P (x).

Using this proposition, we can determine the exact number of irreducible polynomials of
each degree in Zp [x]. For d = 1, P (x) = xp x must have p irreducible factors all of degree
1, which are of course just the polynomials x a for each of the p residue classes a Zp .
2
For d = 2, P (x) = xp x has the p linear factors we just found, together with (p2 p)/2
quadratic factors, since its total degree is p2 . Hence there are (p2 p)/2 distinct monic
irreducible quadratic polynomials over Zp , for every prime p. In the case p = 2, we have
(22 2)/2 = 1. Of the four monic quadratic polynomials in Z2 [x], the unique irreducible one
is x2 + x + 1, since the other three have roots in Z2 .
Continuing in this manner, we find that for d = 3, P (x) must have p linear factors and
3
(p p)/3 factors of degree 3; for d = 4, it must have the p linear factors and (p2 p)/2
quadratic factors that we already discovered, together with (p4 p2 )/4 factors of degree 4,
and so on.
Another, more important, application of the above proposition is to test whether a given
polynomial f (x) Zp [x] is irreducible. Suppose the degree of f (x) is d. If it is not irreducible,
f (x) must have an irreducible factor g(x) of degree at most d/2. Then g(x) is a factor of
e
xp x for some e d/2, so we can discover whether f (x) is irreducible by computing its
e
gcd with each of these polynomials. If f (x) turns out to relatively prime to xp x for all
e d/2, then it is irreducible; otherwise f (x) is reducible.
e
Note that, although the degree pe of xp x might be quite large, the first step in computing
e
e
gcd(f (x), xp x) is to find the remainder of xp x modulo f (x). This remainder is a
polynomial of degree less than d, easily computed by starting with x and taking repeated
p-th powers modulo f (x).
Example. Well test f (x) = x4 + x + 2 for irreducibility in Z3 [x]. It has no root in Z3 [x],
hence no linear factor, so if f (x) is reducible it must be a product of quadratic factors, and
therefore have a common divisor with x9 x (here 9 = p2 ). Modulo f (x) (and reducing all
coefficients modulo 3) we have x4 x + 1, x8 x2 2x + 1 x2 + x + 1, x9 x3 + x2 + x,
and x9 x x3 + x2 . Therefore gcd(f (x), x9 x) = gcd(f (x), x3 + x2 ). Now x3 + x2 factors
as (x + 1) x2 , and we already saw that f (x) has no linear factors, so f (x) is relatively prime
to x3 + x2 . It follows that x4 + x + 2 is irreducible in Z3 [x]. Note that this also implies that
x4 + x + 2 is irreducible in Z[x], and therefore in Q[x], by Gauss Lemma.

Math 113, Summer 2015

Prof. Haiman
Review guide and exercises
1. Outline of topics

Questions on the final exam will cover some subset of the topics listed below. Approximately one half of the exam will be on group theory and basic ring theory. The other half will
be on further ring theory and field theory, which were not covered on the midterm exams.
I have deliberately omitted some topics from the textbook that go beyond what we covered
in class (most of Section 3.5, for example), and some topics that we discussed or will discuss
in class, but are more advanced than what I expect you to know for the exam, such as the
proof of unsolvability of the quintic equation.
Divisibility, prime factorization, GCD for integers. Modular arithmetic. Chinese
Remainder Theorem.
Groups, subgroups, cyclic subgroups, order of an element. Subgroup generated by a
set of elements in a group. Cosets, index of a subgroup, Lagranges theorem. Group
homomorphisms, normal subgroups, quotient groups. Center of a group.
Specific groups: cyclic groups Z and Zn ; group of units Z
n (called (n) in Goodman).
Permutation groups Sn ; even and odd permutations; alternating groups An . Dihedral groups Dn ; rotation groups of regular polyhedra. Groups of invertible matrices
GL(n). Automorphism group Aut(G) of a group.
Partitions and equivalence relations.
Homomorphism theorems for groups.
Direct and semidirect productsexternal construction and internal characterization.
Finitely generated abelian groups: invariant factor decomposition, elementary divisor
decompostion. Use of Smith normal form to compute the invariant factor decompostion of a group presented as Zn /K.
Group actions. Orbits and stablizers. Conjugacy classes and centralizers. Conjugacy
classes in Sn . Burnsides Lemma and its applications. Cauchys theorem. Solvability
of finite p-groups.
Divisibility, factorization into irreducibles, GCD for polynomials.
Rings (commutative rings with identity only) and fields. Subrings. Group of units
in a ring. Direct sum of rings. Ring homomorphisms, ideals, quotient rings. Ideal
generated by a set of elements in a (commutative) ring.
Polynomial rings and evaluation homomorphisms.
Homomorphism theorem and factorization theorem for rings.
Integral domains. Factorization, irreducible and prime elements, units and associates,
GCD in an integral domain. Field of fractions of an integral domain.
PIDs and UFDs. Every PID is a UFD. Gausss Lemma. Polynomial rings over
a UFD are UFDs. Factorization and GCD in a UFD. Rational root test for a
polynomial in R[x] to have a root in Q(R), when R is a UFD.
Prime ideals and maximal ideals. Characterization of prime ideals by R/I being an
integral domain; of maximal ideals by R/I being a field. Implications relating (p)
prime, (p) maximal, p prime, and p irreducible in PIDs and UFDs.
1

 Bases and dimension of finite-dimensional vector spaces over a field.

Dimension of a field extension K L. Finite and algebraic extensions. Formula
dimK (M ) = dimK (L) dimL (M ) for K L M . Minimal polynomial and description of K() for an element algebraic over K. Adjoining algebraic elements to a
field. How to calculate in K().
Characteristic of a field. Existence of finite field F(q) of characteristic p and order
q = pn for every prime p and positive integer n. Multiplicative group F(q) is cyclic.
Automorphism group AutK (L) of a field extension and fixed field Fix(H) of a subgroup H AutK (L).
Existence and uniqueness of splitting fields. Action of AutK (L) on the roots of f (x)
when L is the splitting field of f (x) over K.
Separable polynomials; derivative test.
Galois extensions. Characterization of Galois extensions (Goodman 9.4.15-17). Galois correspondence (Goodman 9.5.4). You should understand the statements of these
theorems and be able to apply them to examples involving fields contained in C.
Note: the definition of a Galois extension K L is that Fix(AutK (L)) = K. For K
L C this is Goodman 7.5.7; for the general case its in the paragraph preceding Theorem
9.4.15.
2. Review exercises
Below are suggested exercises for review. Most of these exercises are similar to the sorts of
questions I might ask on an exam. Some of the multi-part exercises have more parts than an
exam question would have, although the individual parts might be typical of exam questions.
I have also thrown in a few questions that are longer or more difficult than I would put
on an exam, but which serve to illustrate some interesting or important point.
Express the greatest common divisor of 42, 70, and 105 as a linear combination of these
three integers.
Prove that if a2 b2 (mod n), and a 6 b (mod n), then n is composite (i.e., not prime).
Given such an a and b, how can you find a proper factor of n?
Let a, b be elements of a group G, with orders ord(a) = k, ord(b) = l.
(a) Prove that if ab = ba then ord(ab) divides the least common multiple of k and l.
(b) Show that the conclusion of (a) does not have to hold if a and b dont commute, by
finding elements a of order 2 and b of order 3 in S4 such that ab has order 4.
Prove that if G is a group of order 20, then a G satisfies a4 = 1 if and only if a = b5 for
some b G. Hint for only if: what is a5 ?
Show that no two of the groups (Q, +), (Q , ) and (Q>0 , ) are isomorphic. Hint: describe
the elements of finite order and the elements of the form x2 (or 2x if written additively) in
each group.
Let : G H and : H K be group homomorphisms. Prove that the kernel of
is 1 (K), where K = ker().
What are all the elements of the subgroup of Q generated by 2 and 3? Show that this
subgroup is isomorphic to Z Z.

Show that (1 2)(3 4) and (3 4 5) do not generate S5 .

(a) Find the largest conjugacy class in S4 .
(b) Find an element of S4 whose centralizer is as small as possible, and find this centralizer.
In the permutation group S6 , define s = (1 2 3 4 5 6) and t = (1 6)(2 5)(3 4).
(a) Show that tst = s1 .
(b) What is the order of the subgroup hs, ti generated by s and t?
(c) Find an isomorphism between hs, ti and some more familiar group.
(d) How would you generalize the results of this exercise with 6 replaced by any positive
integer n?
Show that the map sending [x]n2 to [x]n is a well-defined, surjective homomorphism from

Z
n2 to Zn .
Show that the map sending [x]n to [1 + nx]n2 is a well-defined, injective homomorphism
from (Zn , +) to Z
n2 , and that its image is equal to the kernel of the homomorphism in the
previous exercise.
Are the rotation groups of the cube and the octahedron isomorphic? Why or why not?
Show that the set SL(n, Z) of n n integer matrices with determinant 1 is a subgroup of
GL(n).
Show that the upper triangular matrices in SL(2, Z) form a subgroup isomorphic to Z2 Z.
(a) Find a group of order 24 in which every element has order 1, 2, 3 or 6.
(b) Find a group of order 24 in which every element has order 1, 2, 3 or 4.
Prove that if g1 H, . . . , gn H are all the distinct left cosets of a subgroup H G, and
h1 K, . . . , hm K are all the distinct left cosets in H of a subgroup K H, then gi hj K are all
the distinct left cosets of K in G. Deduce that if K H G are subgroups, and [G : H]
and [H : K] are finite, then [G : K] = [G : H][H : K], even if G is not a finite group.
Let Z(G) denote the center of G. Prove that if N is a normal subgroup of G, then
Z(G)N/N is contained in the center of G/N . Find an example in which Z(G/N ) is strictly
larger than Z(G)N/N .
Prove that if G = N oK is a semidirect product, and the action of K on N by conjugation
is trivial, then G = N K. In other words, the semidirect product is a direct product in
this case.
(a) Show that multiplication in Zn defines
group automorphisms.
(b) Show that the matrices

a
0

an action : Z
n Aut(Zn ) of Zn on Zn by


b
,
1

where a Z
n and b Zn , form a subgroup of the group of invertible matrices with entries
in Zn .
(c) Show that the semidirect product Zn o Z
n constructed from the action in (a) is
isomorphic to the group of matrices in (b).

Let G be the set of n n real matrices A such that A has exactly one non-zero entry in
every row and column. Let T G be the set of invertible diagonal matrices. Let W G be
the set of matrices with exactly one entry equal to 1 in every row and column, and all other
entries equal to 0.
(a) Show that G is a subgroup of GL(n).
(b) Show that T and W are subgroups of G, and that T is a normal subgroup.
(c) Show that W is isomorphic to Sn .
(d) Show that G is a semidirect product G = T o W .
Let N be the subgroup of Z8 Z12 generated by ([6]8 , [6]12 ). Find a direct product of
cyclic groups isomorphic to (Z8 Z12 )/N .
(a) Find the invariant factor decompostion of Z8 Z2 Z2 Z9 Z3 .
(b) Find the Smith normal form of the diagonal matrix

8 0 0 0 0
0 2 0 0 0

0 0 2 0 0

0 0 0 9 0
0 0 0 0 3
without performing any matrix computations.
Prove that if A is an abelian group of order 20, then A is cyclic if and only if A has an
element of order 4.
Find a chain of normal subgroups
{e} = N0 N1 Nk = D8
such that each Ni /Ni1 is abelian (such a chain must exist, since D8 has order 24 ). What is
the smallest possible value k for the number of steps in such a chain?
Suppose the alternating group A5 acts transitively (i.e., with just one orbit) on a set X
of size |X| = 12. Show that the stabilizer of each x X is a cyclic subgroup generated by a
5-cycle.
How many ways are there to color the vertices of a 10-gon red and blue with 5 of each
color, up to symmetry by rotations in the plane, but not flips?
Goodman Exercise 5.3.7(b)
Compute the gcd of f (x) = 9x3 + 2x 1 and g(x) = 6x2 8x + 2 in Q[x]. Express it as a
linear combination of f (x) and g(x).
Compute the gcd of f (x) = 9x3 + 2x 1 and g(x) = 6x2 8x + 2 in Z[x]. Is it possible
to express it as a linear combination of f (x) and g(x)?
Goodman Exercise 6.2.6
Goodman Exercise 6.5.8
First two sentences of Goodman Exercise 6.5.17
Goodman Exercise 6.5.21

Prove that x3 y 2xy + x5 is irreducible in R[x, y]. Hint: observe that it is irreducible in
R(x)[y].
Let a and b be non-zero elements of a UFD R, and let d be a gcd of a and b. Show that
m = ab/d is a least common multiple of a and b. That is, both a and b divide m, and m
divides every common multiple of a and b.
Note that the expression ab/d denotes an element of the fraction field of R, but since d
divides ab, this element is actually in R.
Prove that if f and g are relatively prime elements of a UFD R, then the intersection of
the principal ideals (f ) and (g) is equal to (f g).
Show that if a and b are elements of a field K, and a 6= b, then the ring K[x]/((xa)(xb))
is isomorphic to K K. Start by finding a homomorphism from K[x] to K K whose kernel
is ((x a)(x b)).
Show that the condition a 6= b in the previous exercise cannot be omitted, by proving
that the rings K[x]/((x a)2 ) and K K are not isomorphic. Hint: consider elements r
satisfying r2 = 0 in each ring.
Let : Q[x, y] Q[t] be the evaluation homomorphism p(x, y) 7 p(t2 , t3 ).
(a) Show that the image S of consists of all polynomials f (t) in which t1 has coefficient
zero. In particular, this set S is a subring of Q[t].
(b) Show that the ideal (y 2 x3 ) is contained in the kernel of , and use this to define a
surjective homomorphism : Q[x, y]/(y 2 x3 ) S
(c) Show that every element of Q[x, y]/(y 2 x3 ) can be expressed in the form a(x) + yb(x)
(more precisely, as the congruence class of a(x) + yb(x)).
(d) Show that is injective, and therefore Q[x, y]/(y 2 x3 )
= S.
2
3
(e) Deduce that (y x ) is a prime ideal in Q[x, y].
Prove that (x2 2, y 1) is a maximal ideal in Q[x, y].
(a) Show that x3 + 2x + 2 is irreducible in Q[x] and has only one real root.
(b) Let L = Q[x]/(x3 + 2x + 2). Show that L is a field isomorphic to Q(), where is the
real root of x3 + 2x + 2.
(c) Show that AutQ (L) is the trivial group. In particular, L is not a Galois extension of
Q.
There are eight monic polynomials of degree 4 over Z2 , of which three are irreducible.
(a) Find the irreducible ones by eliminating the five which factor.
(b) Since F(16) is isomorphic to Z15 , there are (15) = 8 elements F(16) such that
has order 15 in F(16) , i.e., such that generates F(16) as a cyclic group. These eight
elements must be the roots of two of the degree 4 irreducible polynomials in (a) (four roots
each), with the roots of the third one having order less than 15 in F(16) .
Which one of the three irreducible polyomials in (a) has roots of order less than 15 and
what is their order in F(16) ?
(a) Show that no expression involving only rational numbers, arithmetric operations
(ad
3
dition, subtraction, multiplication
and division) and square roots can be equal to 2.

n
(b) Show the same for 2 if n is not a power of two.

Goodman Exercise 7.3.11

Goodman Exercise 7.4.3(b).
Let = e2i/5 .
(a) Show that Q Q(cos 2/5) Q() and that no two of these fields are equal. (To
show that cos 2/5 is irrational, find its minimal polynomial and show that it is irreducible
over Q.)
(b) Deduce from (a) that the minimal polynomial of over Q has degree at least 4.
(c) Show that is a root of f (x) = x4 + x3 + x2 + x + 1. Hint: use x5 1 = (x 1)(x4 +
x3 + x2 + x + 1). Deduce that f (x) is the minimal polynomial of .
(d) Find all the complex roots of f (x) and show that Q() is its splitting field.
(e) Determine the Galois group AutQ (Q()) and its action on the roots of f (x).

Let L = Q( 2, 3)
(a) Show that L is the splitting field over Q of f (x) = (x2 2)(x2 3), hence Q L is a
Galois extension.
(b) Find a basis of L over Q and give the rule for multiplying two elements of L expressed
as linear combinations of the basis elements.
(c) Find the Galois group AutQ (L) and describe its action on the roots of f (x).
(d) Find all intermediate fields Q E
L.
(e) Find the AutQ (L) orbit of = 2 + 3.
(f) Use (e) to find the minimal polynomial of over Q
(g) Deduce from (e) or (f) that L = Q().
Let Q = Zp (u) be the field of rational functions in one variable u over Zp , that is, the
fraction field of Zp [u].
(a) Show that f (x) = xp u is irreducible in Zp [u, x], and therefore also in Q[x] = Zp (u)[x]
by Gausss Lemma.
(b) Show that in Q(u1/p ) = Q[x]/(f (x)), the element u1/p is a root of f (x) of multiplicity
p, that is, f (x) = (x u1/p )p . Deduce that Q(u1/p ) is the splitting field of f (x) over Q, even
though f (x) has only one root in this field. In particular, Q(u1/p ) is not a Galois extension
of Q.
(c) Part (b) implies that f (x) is not a separable polynomial over Q. Verify that the
derivative test also shows this.